[PATCH] Light weight event counters

The remaining counters in page_state after the zoned VM counter patches
have been applied are all just for show in /proc/vmstat.  They have no
essential function for the VM.

We use a simple increment of per cpu variables.  In order to avoid the most
severe races we disable preempt.  Preempt does not prevent the race between
an increment and an interrupt handler incrementing the same statistics
counter.  However, that race is exceedingly rare, we may only loose one
increment or so and there is no requirement (at least not in kernel) that
the vm event counters have to be accurate.

In the non preempt case this results in a simple increment for each
counter.  For many architectures this will be reduced by the compiler to a
single instruction.  This single instruction is atomic for i386 and x86_64.
 And therefore even the rare race condition in an interrupt is avoided for
both architectures in most cases.

The patchset also adds an off switch for embedded systems that allows a
building of linux kernels without these counters.

The implementation of these counters is through inline code that hopefully
results in only a single instruction increment instruction being emitted
(i386, x86_64) or in the increment being hidden though instruction
concurrency (EPIC architectures such as ia64 can get that done).

Benefits:
- VM event counter operations usually reduce to a single inline instruction
  on i386 and x86_64.
- No interrupt disable, only preempt disable for the preempt case.
  Preempt disable can also be avoided by moving the counter into a spinlock.
- Handling is similar to zoned VM counters.
- Simple and easily extendable.
- Can be omitted to reduce memory use for embedded use.

References:

RFC http://marc.theaimsgroup.com/?l=linux-kernel&m=113512330605497&w=2
RFC http://marc.theaimsgroup.com/?l=linux-kernel&m=114988082814934&w=2
local_t http://marc.theaimsgroup.com/?l=linux-kernel&m=114991748606690&w=2
V2 http://marc.theaimsgroup.com/?t=115014808400007&r=1&w=2
V3 http://marc.theaimsgroup.com/?l=linux-kernel&m=115024767022346&w=2
V4 http://marc.theaimsgroup.com/?l=linux-kernel&m=115047968808926&w=2

Signed-off-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

arch/s390/appldata/appldata_base.c

@@ -766,7 +766,6 @@ unsigned long nr_iowait(void)
 #endif /* MODULE */
 EXPORT_SYMBOL_GPL(si_swapinfo);
 EXPORT_SYMBOL_GPL(nr_threads);
-EXPORT_SYMBOL_GPL(get_full_page_state);
 EXPORT_SYMBOL_GPL(nr_running);
 EXPORT_SYMBOL_GPL(nr_iowait);
 //EXPORT_SYMBOL_GPL(nr_context_switches);

arch/s390/appldata/appldata_mem.c

@@ -107,21 +107,21 @@ static void appldata_get_mem_data(void *data)
 	 * serialized through the appldata_ops_lock and can use static
 	 */
 	static struct sysinfo val;
-	static struct page_state ps;
+	unsigned long ev[NR_VM_EVENT_ITEMS];
 	struct appldata_mem_data *mem_data;
 
 	mem_data = data;
 	mem_data->sync_count_1++;
 
-	get_full_page_state(&ps);
-	mem_data->pgpgin     = ps.pgpgin >> 1;
-	mem_data->pgpgout    = ps.pgpgout >> 1;
-	mem_data->pswpin     = ps.pswpin;
-	mem_data->pswpout    = ps.pswpout;
-	mem_data->pgalloc    = ps.pgalloc_high + ps.pgalloc_normal +
-			       ps.pgalloc_dma;
-	mem_data->pgfault    = ps.pgfault;
-	mem_data->pgmajfault = ps.pgmajfault;
+	all_vm_events(ev);
+	mem_data->pgpgin     = ev[PGPGIN] >> 1;
+	mem_data->pgpgout    = ev[PGPGOUT] >> 1;
+	mem_data->pswpin     = ev[PSWPIN];
+	mem_data->pswpout    = ev[PSWPOUT];
+	mem_data->pgalloc    = ev[PGALLOC_HIGH] + ev[PGALLOC_NORMAL] +
+			       ev[PGALLOC_DMA];
+	mem_data->pgfault    = ev[PGFAULT];
+	mem_data->pgmajfault = ev[PGMAJFAULT];
 
 	si_meminfo(&val);
 	mem_data->sharedram = val.sharedram;

block/ll_rw_blk.c

@@ -3117,9 +3117,9 @@ void submit_bio(int rw, struct bio *bio)
 	BIO_BUG_ON(!bio->bi_io_vec);
 	bio->bi_rw |= rw;
 	if (rw & WRITE)
-		mod_page_state(pgpgout, count);
+		count_vm_events(PGPGOUT, count);
 	else
-		mod_page_state(pgpgin, count);
+		count_vm_events(PGPGIN, count);
 
 	if (unlikely(block_dump)) {
 		char b[BDEVNAME_SIZE];

drivers/parisc/led.c

@@ -411,16 +411,17 @@ static __inline__ int led_get_net_activity(void)
 static __inline__ int led_get_diskio_activity(void)
 {	
 	static unsigned long last_pgpgin, last_pgpgout;
-	struct page_state pgstat;
+	unsigned long events[NR_VM_EVENT_ITEMS];
 	int changed;
 
-	get_full_page_state(&pgstat); /* get no of sectors in & out */
+	all_vm_events(events);
 
 	/* Just use a very simple calculation here. Do not care about overflow,
 	   since we only want to know if there was activity or not. */
-	changed = (pgstat.pgpgin != last_pgpgin) || (pgstat.pgpgout != last_pgpgout);
-	last_pgpgin  = pgstat.pgpgin;
-	last_pgpgout = pgstat.pgpgout;
+	changed = (events[PGPGIN] != last_pgpgin) ||
+		  (events[PGPGOUT] != last_pgpgout);
+	last_pgpgin  = events[PGPGIN];
+	last_pgpgout = events[PGPGOUT];
 
 	return (changed ? LED_DISK_IO : 0);
 }

fs/inode.c

@@ -452,15 +452,14 @@ static void prune_icache(int nr_to_scan)
 		nr_pruned++;
 	}
 	inodes_stat.nr_unused -= nr_pruned;
+	if (current_is_kswapd())
+		__count_vm_events(KSWAPD_INODESTEAL, reap);
+	else
+		__count_vm_events(PGINODESTEAL, reap);
 	spin_unlock(&inode_lock);
 
 	dispose_list(&freeable);
 	mutex_unlock(&iprune_mutex);
-
-	if (current_is_kswapd())
-		mod_page_state(kswapd_inodesteal, reap);
-	else
-		mod_page_state(pginodesteal, reap);
 }
 
 /*

fs/ncpfs/mmap.c

@@ -93,7 +93,7 @@ static struct page* ncp_file_mmap_nopage(struct vm_area_struct *area,
 	 */
 	if (type)
 		*type = VM_FAULT_MAJOR;
-	inc_page_state(pgmajfault);
+	count_vm_event(PGMAJFAULT);
 	return page;
 }
 

include/linux/vmstat.h

@@ -7,115 +7,77 @@
 #include <linux/mmzone.h>
 #include <asm/atomic.h>
 
+#ifdef CONFIG_VM_EVENT_COUNTERS
 /*
- * Global page accounting.  One instance per CPU.  Only unsigned longs are
- * allowed.
+ * Light weight per cpu counter implementation.
  *
- * - Fields can be modified with xxx_page_state and xxx_page_state_zone at
- * any time safely (which protects the instance from modification by
- * interrupt.
- * - The __xxx_page_state variants can be used safely when interrupts are
- * disabled.
- * - The __xxx_page_state variants can be used if the field is only
- * modified from process context and protected from preemption, or only
- * modified from interrupt context.  In this case, the field should be
- * commented here.
+ * Counters should only be incremented and no critical kernel component
+ * should rely on the counter values.
+ *
+ * Counters are handled completely inline. On many platforms the code
+ * generated will simply be the increment of a global address.
  */
-struct page_state {
-	unsigned long pgpgin;		/* Disk reads */
-	unsigned long pgpgout;		/* Disk writes */
-	unsigned long pswpin;		/* swap reads */
-	unsigned long pswpout;		/* swap writes */
-
-	unsigned long pgalloc_high;	/* page allocations */
-	unsigned long pgalloc_normal;
-	unsigned long pgalloc_dma32;
-	unsigned long pgalloc_dma;
-
-	unsigned long pgfree;		/* page freeings */
-	unsigned long pgactivate;	/* pages moved inactive->active */
-	unsigned long pgdeactivate;	/* pages moved active->inactive */
-
-	unsigned long pgfault;		/* faults (major+minor) */
-	unsigned long pgmajfault;	/* faults (major only) */
-
-	unsigned long pgrefill_high;	/* inspected in refill_inactive_zone */
-	unsigned long pgrefill_normal;
-	unsigned long pgrefill_dma32;
-	unsigned long pgrefill_dma;
-
-	unsigned long pgsteal_high;	/* total highmem pages reclaimed */
-	unsigned long pgsteal_normal;
-	unsigned long pgsteal_dma32;
-	unsigned long pgsteal_dma;
-
-	unsigned long pgscan_kswapd_high;/* total highmem pages scanned */
-	unsigned long pgscan_kswapd_normal;
-	unsigned long pgscan_kswapd_dma32;
-	unsigned long pgscan_kswapd_dma;
-
-	unsigned long pgscan_direct_high;/* total highmem pages scanned */
-	unsigned long pgscan_direct_normal;
-	unsigned long pgscan_direct_dma32;
-	unsigned long pgscan_direct_dma;
-
-	unsigned long pginodesteal;	/* pages reclaimed via inode freeing */
-	unsigned long slabs_scanned;	/* slab objects scanned */
-	unsigned long kswapd_steal;	/* pages reclaimed by kswapd */
-	unsigned long kswapd_inodesteal;/* reclaimed via kswapd inode freeing */
-	unsigned long pageoutrun;	/* kswapd's calls to page reclaim */
-	unsigned long allocstall;	/* direct reclaim calls */
-
-	unsigned long pgrotated;	/* pages rotated to tail of the LRU */
+
+#define FOR_ALL_ZONES(x) x##_DMA, x##_DMA32, x##_NORMAL, x##_HIGH
+
+enum vm_event_item { PGPGIN, PGPGOUT, PSWPIN, PSWPOUT,
+		FOR_ALL_ZONES(PGALLOC),
+		PGFREE, PGACTIVATE, PGDEACTIVATE,
+		PGFAULT, PGMAJFAULT,
+		FOR_ALL_ZONES(PGREFILL),
+		FOR_ALL_ZONES(PGSTEAL),
+		FOR_ALL_ZONES(PGSCAN_KSWAPD),
+		FOR_ALL_ZONES(PGSCAN_DIRECT),
+		PGINODESTEAL, SLABS_SCANNED, KSWAPD_STEAL, KSWAPD_INODESTEAL,
+		PAGEOUTRUN, ALLOCSTALL, PGROTATED,
+		NR_VM_EVENT_ITEMS
+};
+
+struct vm_event_state {
+	unsigned long event[NR_VM_EVENT_ITEMS];
 };
 
-extern void get_full_page_state(struct page_state *ret);
-extern void mod_page_state_offset(unsigned long offset, unsigned long delta);
-extern void __mod_page_state_offset(unsigned long offset, unsigned long delta);
-
-#define mod_page_state(member, delta)	\
-	mod_page_state_offset(offsetof(struct page_state, member), (delta))
-
-#define __mod_page_state(member, delta)	\
-	__mod_page_state_offset(offsetof(struct page_state, member), (delta))
-
-#define inc_page_state(member)		mod_page_state(member, 1UL)
-#define dec_page_state(member)		mod_page_state(member, 0UL - 1)
-#define add_page_state(member,delta)	mod_page_state(member, (delta))
-#define sub_page_state(member,delta)	mod_page_state(member, 0UL - (delta))
-
-#define __inc_page_state(member)	__mod_page_state(member, 1UL)
-#define __dec_page_state(member)	__mod_page_state(member, 0UL - 1)
-#define __add_page_state(member,delta)	__mod_page_state(member, (delta))
-#define __sub_page_state(member,delta)	__mod_page_state(member, 0UL - (delta))
-
-#define page_state(member) (*__page_state(offsetof(struct page_state, member)))
-
-#define state_zone_offset(zone, member)					\
-({									\
-	unsigned offset;						\
-	if (is_highmem(zone))						\
-		offset = offsetof(struct page_state, member##_high);	\
-	else if (is_normal(zone))					\
-		offset = offsetof(struct page_state, member##_normal);	\
-	else if (is_dma32(zone))					\
-		offset = offsetof(struct page_state, member##_dma32);	\
-	else								\
-		offset = offsetof(struct page_state, member##_dma);	\
-	offset;								\
-})
-
-#define __mod_page_state_zone(zone, member, delta)			\
- do {									\
-	__mod_page_state_offset(state_zone_offset(zone, member), (delta)); \
- } while (0)
-
-#define mod_page_state_zone(zone, member, delta)			\
- do {									\
-	mod_page_state_offset(state_zone_offset(zone, member), (delta)); \
- } while (0)
-
-DECLARE_PER_CPU(struct page_state, page_states);
+DECLARE_PER_CPU(struct vm_event_state, vm_event_states);
+
+static inline void __count_vm_event(enum vm_event_item item)
+{
+	__get_cpu_var(vm_event_states.event[item])++;
+}
+
+static inline void count_vm_event(enum vm_event_item item)
+{
+	get_cpu_var(vm_event_states.event[item])++;
+	put_cpu();
+}
+
+static inline void __count_vm_events(enum vm_event_item item, long delta)
+{
+	__get_cpu_var(vm_event_states.event[item]) += delta;
+}
+
+static inline void count_vm_events(enum vm_event_item item, long delta)
+{
+	get_cpu_var(vm_event_states.event[item])++;
+	put_cpu();
+}
+
+extern void all_vm_events(unsigned long *);
+extern void vm_events_fold_cpu(int cpu);
+
+#else
+
+/* Disable counters */
+#define get_cpu_vm_events(e)	0L
+#define count_vm_event(e)	do { } while (0)
+#define count_vm_events(e,d)	do { } while (0)
+#define __count_vm_event(e)	do { } while (0)
+#define __count_vm_events(e,d)	do { } while (0)
+#define vm_events_fold_cpu(x)	do { } while (0)
+
+#endif /* CONFIG_VM_EVENT_COUNTERS */
+
+#define __count_zone_vm_events(item, zone, delta) \
+			__count_vm_events(item##_DMA + zone_idx(zone), delta)
 
 /*
  * Zone based page accounting with per cpu differentials.

init/Kconfig

@@ -379,6 +379,15 @@ config SLAB
 	  SLOB is more space efficient but does not scale well and is
 	  more susceptible to fragmentation.
 
+config VM_EVENT_COUNTERS
+	default y
+	bool "Enable VM event counters for /proc/vmstat" if EMBEDDED
+	help
+	  VM event counters are only needed to for event counts to be
+	  shown. They have no function for the kernel itself. This
+	  option allows the disabling of the VM event counters.
+	  /proc/vmstat will only show page counts.
+
 endmenu		# General setup
 
 config TINY_SHMEM

mm/filemap.c

@@ -1416,7 +1416,7 @@ struct page *filemap_nopage(struct vm_area_struct *area,
 		 */
 		if (!did_readaround) {
 			majmin = VM_FAULT_MAJOR;
-			inc_page_state(pgmajfault);
+			count_vm_event(PGMAJFAULT);
 		}
 		did_readaround = 1;
 		ra_pages = max_sane_readahead(file->f_ra.ra_pages);
@@ -1487,7 +1487,7 @@ struct page *filemap_nopage(struct vm_area_struct *area,
 page_not_uptodate:
 	if (!did_readaround) {
 		majmin = VM_FAULT_MAJOR;
-		inc_page_state(pgmajfault);
+		count_vm_event(PGMAJFAULT);
 	}
 	lock_page(page);
 

mm/memory.c

@@ -1951,7 +1951,7 @@ static int do_swap_page(struct mm_struct *mm, struct vm_area_struct *vma,
 
 		/* Had to read the page from swap area: Major fault */
 		ret = VM_FAULT_MAJOR;
-		inc_page_state(pgmajfault);
+		count_vm_event(PGMAJFAULT);
 		grab_swap_token();
 	}
 
@@ -2324,7 +2324,7 @@ int __handle_mm_fault(struct mm_struct *mm, struct vm_area_struct *vma,
 
 	__set_current_state(TASK_RUNNING);
 
-	inc_page_state(pgfault);
+	count_vm_event(PGFAULT);
 
 	if (unlikely(is_vm_hugetlb_page(vma)))
 		return hugetlb_fault(mm, vma, address, write_access);

mm/page_alloc.c

@@ -456,7 +456,7 @@ static void __free_pages_ok(struct page *page, unsigned int order)
 
 	kernel_map_pages(page, 1 << order, 0);
 	local_irq_save(flags);
-	__mod_page_state(pgfree, 1 << order);
+	__count_vm_events(PGFREE, 1 << order);
 	free_one_page(page_zone(page), page, order);
 	local_irq_restore(flags);
 }
@@ -729,7 +729,7 @@ static void fastcall free_hot_cold_page(struct page *page, int cold)
 
 	pcp = &zone_pcp(zone, get_cpu())->pcp[cold];
 	local_irq_save(flags);
-	__inc_page_state(pgfree);
+	__count_vm_event(PGFREE);
 	list_add(&page->lru, &pcp->list);
 	pcp->count++;
 	if (pcp->count >= pcp->high) {
@@ -805,7 +805,7 @@ static struct page *buffered_rmqueue(struct zonelist *zonelist,
 			goto failed;
 	}
 
-	__mod_page_state_zone(zone, pgalloc, 1 << order);
+	__count_zone_vm_events(PGALLOC, zone, 1 << order);
 	zone_statistics(zonelist, zone);
 	local_irq_restore(flags);
 	put_cpu();
@@ -2101,24 +2101,11 @@ static int page_alloc_cpu_notify(struct notifier_block *self,
 				 unsigned long action, void *hcpu)
 {
 	int cpu = (unsigned long)hcpu;
-	unsigned long *src, *dest;
 
 	if (action == CPU_DEAD) {
-		int i;
-
 		local_irq_disable();
 		__drain_pages(cpu);
-
-		/* Add dead cpu's page_states to our own. */
-		dest = (unsigned long *)&__get_cpu_var(page_states);
-		src = (unsigned long *)&per_cpu(page_states, cpu);
-
-		for (i = 0; i < sizeof(struct page_state)/sizeof(unsigned long);
-				i++) {
-			dest[i] += src[i];
-			src[i] = 0;
-		}
-
+		vm_events_fold_cpu(cpu);
 		local_irq_enable();
 		refresh_cpu_vm_stats(cpu);
 	}